Sutskever's New Company, SSI, Bets on Google Cloud TPUs to Accelerate Safe Superintelligent AI Development

AIbase基地

Published inAI News · 4 min read · Apr 10, 2025

Safe Superintelligence (SSI), a new AI startup founded by Ilya Sutskever, co-founder and former chief scientist of OpenAI, has announced a partnership with Google Cloud. According to a press release issued on Wednesday, SSI will leverage Google Cloud's TPU chips to accelerate the development of its "safe superintelligence AI." Google Cloud revealed that SSI is using its TPUs to boost R&D speed in building safe superintelligent AI.

AI Robot Interview and Negotiation

Image Source Note: Image generated by AI, licensed through Midjourney

Multiple cloud providers are vying for the business of a small number of unicorn AI startups that spend hundreds of millions of dollars annually training AI foundation models. Sources told TechCrunch that this partnership signifies SSI will allocate a significant portion of its computing budget to Google Cloud, making it SSI's primary computing provider. Notably, Google Cloud has previously struck computing deals with several former AI researchers, including Sutskever, many of whom now run multi-billion dollar AI startups. For example, last October, Google Cloud announced it was the primary computing provider for World Labs, founded by former Google Cloud AI chief scientist Fei-Fei Li.

It remains unclear whether SSI has partnerships with other cloud or computing providers. Spokespeople for both Google Cloud and Safe Superintelligence declined to comment.

SSI was officially founded in June 2024, coinciding with Sutskever's departure as OpenAI's chief scientist. The company has secured a substantial $1 billion in funding from prominent venture capital firms including Andreessen Horowitz, Sequoia Capital, DST Global, and SV Angel.

Since its inception, SSI has maintained a low profile regarding its specific activities. Its website states that its core mission is the development of safe superintelligent AI systems, and this is its sole focus. Sutskever previously stated that he has identified new research directions focused on improving the performance of cutting-edge AI models.

Before co-founding OpenAI, Sutskever spent years at Google Brain researching neural networks. After years leading OpenAI's AI safety efforts, he played a key role in the November 2023 departure of OpenAI CEO Sam Altman and subsequently participated in the employee movement to reinstate him. Following this turmoil, Sutskever reportedly stayed away from the OpenAI office for several months before ultimately leaving to found SSI.

AI Daily: OpenAI to Potentially Release GPT-4.1 Series Next Week; Pika's New AI Video Feature 'Twists'; SenseTime's 'SenseNova' V6 Makes a Stunning Debut

Welcome to the AI Daily column! Your daily guide to exploring the world of artificial intelligence. We present you with the hottest content in the AI field, focusing on developers and helping you understand technology trends and innovative AI product applications. Discover new AI products here: https://top.aibase.com/ 1. Reports suggest OpenAI will release the GPT-4.1 series next week, including Mini and Nano versions. OpenAI's upcoming release of the GPT-4.1 and o3 series marks a significant advancement in...

Report: OpenAI to Release GPT-4.1 Series Next Week, Including Mini and Nano Versions

AI leader OpenAI is poised to unleash a new wave of technological advancements next week! According to tech media outlet The Verge, OpenAI plans to launch a major update including the GPT-4.1 series, o3 series, and several other AI models. This flurry of releases not only demonstrates OpenAI's ambition for accelerated innovation but also provides the industry with more powerful AI tools. GPT-4.1 Series: A Comprehensive Upgrade in Multimodal Capabilities As the successor to GPT-4.0, the GPT-4.1 series...

OpenAI Open-Sources BrowseComp: A New Benchmark for Evaluating AI Agent Web Browsing Capabilities

A new benchmark for evaluating AI agents has arrived! OpenAI has announced the open-sourcing of BrowseComp, an innovative benchmark designed specifically to assess the web browsing capabilities of AI agents. This initiative provides the AI research community with a new tool and lays the foundation for more intelligent and reliable browsing agents. AIbase offers an in-depth analysis of BrowseComp's core value and industry impact. BrowseComp: The ultimate test for AI browsing capabilities.

ChatGPT Launches Long-Term Memory Feature: A New Era for AI Interaction

OpenAI has announced a major update: ChatGPT now officially features long-term memory! This is considered one of the most significant upgrades since ChatGPT's launch, promising a greatly enhanced user experience and ushering in a new era of personalized interaction. AIbase provides an exclusive breakdown of this feature's key highlights and potential impact. While ChatGPT has long been a productivity tool for many users thanks to its powerful language processing capabilities, its memory has been limited to single conversations or short-term contexts. Now, with this new feature...

Former OpenAI CTO's Startup Adds Key OpenAI Alumni

Thinking Machines Lab, the startup founded by former OpenAI CTO Mira Murati, has added two heavyweight advisors from her former employer: former OpenAI Chief Scientist Bob McGrew and former OpenAI researcher Alec Radford. Their contributions bring significant expertise and energy to the company, which is focused on developing AI that meets individual needs.

OpenAI's New Image Generator Sparks Controversy; CEO Altman Rebuts Miyazaki's Criticism

OpenAI recently released a new image generator that has drawn attention for its ability to produce illustrations mimicking the style of Studio Ghibli, but criticized by some users for lacking "soul." Meanwhile, OpenAI CEO Sam Altman has fiercely rebutted critics, including Studio Ghibli co-founder Hayao Miyazaki. Altman responded to Miyazaki's harsh criticism during an interview with tech founder and YouTuber Arun Mayya, openly refuting...

How Publishers Can Monetize AI: Exploring Revenue Models and Their Pros and Cons

With the rapid advancement of AI technology, publishers are seeking new revenue streams to adapt to this evolving market. Recently, major publishers have been forging partnerships with AI companies, exploring diverse revenue models including content licensing agreements and ad revenue sharing. These new business models offer publishers potential profit opportunities but also present complex relationships with AI companies. Content licensing agreements are one of the most common collaborations between publishers and AI companies. Publishers license their content to AI companies...